Denafrips R2R DAC (plays 512 DSD) reviewed as giant killer

[Miska]

From today's point of view, Loki DAC is also a bit limited with only DSD64 support. This is likely mainly due to limitation of the USB interface Schiit uses (Cmedia) across the product range which is limited to max 192 kHz PCM, which in turn limits DoP to DSD64.

 

For the price at the time when it was introduced it was a fine DSD DAC. From today's point of view, it is just outdated.

 

[Miska]

On 8/26/2018 at 8:09 PM, Reggy said:

I don't know how you guys get to enjoy it with on the fly conversion but for me its a gimmick that suppose to sound better but in the end it makes no real world difference except for added CPU usage and fan noise. There's been plenty of papers on the PCM Vs. DSD. , in the end most people would justify that PCM sounds more immediate while DSD is more rounded in sound or soft. 

 

I use either fanless machines, or machines that have quiet enough fans that I cannot hear them even at full CPU load. I hate hearing fan noise, regardless if it is music playback or processing a video with video editor. Or of course you can just have the machine with noisy fans located somewhere else in the house.

 

On 8/26/2018 at 8:09 PM, Reggy said:

If it were me I would say focus on the conversion process or design better chips to be able to avoid having to resort to DSD. I think the R2R DACs are starting to sprawl and do it right.

 

Better chips would make the chips more expensive, while chips are designed to a certain price point. And likely have other design requirements such as no cooling solution needed. Also the very close proximity of sensitive analog parts on the same die impose limits on how much DSP processing power can be put on the die.

 

R2R DACs have their own inherent restrictions and so far I haven't seen a device that would have overcome these. This is why delta-sigma (including DSD) DACs have taken over. And with R2R you also need to use good digital filters and high as possible conversion rate to minimize the conversion errors.

 

Anyway, DACs like Denafrips and Holo Audio are best for comparing PCM (R2R) vs SDM (DSD) DACs, because there you have two native DACs everything else being the same. Of course these are still subject to particular implementation of the actual conversion.

[Miska]

On 8/26/2018 at 6:19 PM, giordy60 said:

Interesting .......
I would like to know the opinion of Miska (since he has developed the DSC1) .....
( if he has had an ares in his hands ) 
or in denafrips would explain better what the little jewel of the house

?

 

I'm very interested on Denafrips DACs (and all other discrete implementations). They look very much similar to Holo Audio DACs. I'm pondering between Ares and Pontus...

 

[Miska]

52 minutes ago, Reggy said:

I do have the Ares and I am enjoying the musical delivery and presentation very much on my Genelec's. What would the purpose of you considering buying one of these DACs? A side grade, something different to listen to? Would you be using headphones, speakers, studio monitors?

 

I don't yet have a discrete DAC in my living room system, and Ares would fit on the small shelf there. I would like to have a discrete DAC in all of my systems. Currently I have Holo Audio Spring DAC L2 and T+A DAC8 DSD in the listening room and Holo Audio Cyan DSD on my office desk headphone system. Of course I have big bunch of other DACs too, but I like to have a discrete choice available in all systems.

 

But other than that, I'd like to measure/test it with HQPlayer just like all the other DACs too. Every additional DAC improves QA coverage...

 

[Miska]

2 minutes ago, Reggy said:

Then again, having said that I think getting the Venus would be a benefit for the community because it hasn't been measured and it is in the same price range as the Schitt Yggydrasil, I haven't heard it but I believe it would be a much stronger upgrade over the Ares, and the Pontus. Unfortunately there is no coverage on that mid-tier product.

 

Fancier would be nicer, but at ~2400€ vs ~1400€ of Pontus there's quite a bit of price difference. Technically from Pontus to Terminator these seem to quite similar architecture, so I wonder if it would make any big difference in measurement results.

 

[Miska]

1 minute ago, Reggy said:

           True from Ares to Pontus there are differences however there are also major differences from Pontus to Venus, while Terminator the differences are a bit minimal compared to Venus.

 

With the Venus there are the 0.005% precision resistors vs 0.01% for the Pontus and the added Femto clocks.Whether there are measurable differences, I suspect so.

 

With the analog FIR type DSD DAC the resistor precision doesn't really matter much. And R2R beyond 20-bits certainly doesn't make any difference. The clocks could make a difference, but paying 1000€ extra to get 25€ clock component instead of let's say 10€ clock component is not something I'm ready to do. And usually clocks are spoiled so much more elsewhere that what I've measured those "femto" clocks don't necessarily make any real difference in actual output.

 

8 minutes ago, Reggy said:

Torq said: "On detail and resolution, I feel that the PONTUS, while meaningfully ahead of the ARES here, is not quite on the same level as something like Gungnir MB, the Spring DAC L1 and definitely not where the dac1541 exhibits its biggest strength. This is more noticeable using dynamic cans (Utopia, HD800S) than it was with planar units (LCD-4, Abyss, Abyss Phi). I wouldn’t go as far as to say that detail is being “smeared over”, but the unit is not resolving at the quite same level as its principal competition.

 

I wonder what was used to do oversampling in this case and whether it was comparing PCM or DSD side of the DAC.

 

 

[Miska]

1 hour ago, barrows said:

Most solid state gear will last a very long time when kept powered up 24/7.  The possible exception being class A amplifiers which run very hot (the heat can dry out electrolytic capacitors).

A DAC should be left on at all times for best performance, clocks require quite a bit of warm up to stabilize at their lowest noise level.  Indeed, many power up/down cycles put more stress on solid state components than leaving them on, as the inrush currents at power up are many times higher than at any other time.  Leave it powered dup all the time, except when away on vacation.

 

I'm turning off my gear when not in use. I'm a bit worried about electrolytic capacitors, I've had quite many blown up after couple of years running 24/7. If the caps in hardware are rated for 105C or hotter, then they'll last a bit bitter. But standard ones are rated for 85C max. Heat almost exponentially reduces electrolytic capacitor lifetime, so it really depends on what kind of internal temperatures it reaches and what kind of components have been selected....

 

1 hour ago, Ralf11 said:

or when a thunderstorm is about to pepper the area with lightening strikes

 

That's another thing. Although I use powerline filters and surge protectors. And Finland is not particularly bad in that respect, especially now that almost all electrical cabling has been moved underground due to legal reliability requirements.

 

[Miska]

44 minutes ago, JamesBardsley said:

I actually only asked about servicing as most hifi equipment that contains capacitors will tend to need servicing after say 10 to 15 years as the caps eventually dry out and get replaced.

 

Good caps shouldn't go bad in 10 years unless you keep the device powered up all the time or unused for years...

 

[Miska]

10 minutes ago, JamesBardsley said:

the caps dry out inside and need replacing after around 10-15 years - the details depend on the temperature of room, equipment, etc.

That's normal as far as I know.

Thermodynamics dicates it as a heat driven process...

 

Electrolytic caps have limited number of power-on usage hours. A bit like tubes too. So after some number of usage hours they just fail.

 

I have lot of gear, even 30 years old that still runs fine without ever replacing caps. And I always turn it off when not in use..

 

Caps "dry out" primarily if not used for a very long time. If being unused for 10 years or so, it would be good to give the caps soft start through resistor to let the layers rebuild properties.

 

[Miska]

3 hours ago, austinpop said:

The Chord dual BNC mode requires support on both ends, as in this mode, the M Scaler splits L and R channels onto the 2 AES inputs.

 

At least HQPlayer supports dual-wire output mode, if the hardware itself doesn't do the splitting...

 

[Miska]

On 6/3/2019 at 12:56 AM, JamesBardsley said:

I shall always worry about EM noise getting into my hifi, as that's one of the main causes of bad SQ

and controlling EM noise is therefore one of the best ways to get better SQ

 

I don't use a PC for hifi at all

 

my signal chain is Tidal - Plusnet router - Cisco Catalyst 2960 switch (not currently in use) - dCS Network Bridge - Naim 272 (streamer/DAC/preamp) - Naim 150x power amp -Naim Ariva Speakers

 

If you have an isolation layer like network between PC and DAC, you don't have a problem.

 

In your current setup you do have a PC - Tidal is running on pile of PC's located in datacenters across the globe. (IIRC, last I checked they were using Cloudflare as CDN provider)

[Miska]

22 hours ago, JamesBardsley said:

Yup, tidal servers are my nas.

thats why I’m working on how I can min8mise the em mess created in my isp plus-net router ... 

 

Have the Cisco switch between the router and rest and you are done...  Just remember to check that you have 802.3x flow control enabled in the Cisco settings.

 

(and remember, no shielded ethernet cables!)

 

[Miska]

5 hours ago, Paul R said:

At $150, you should be able to buy a desktop drive that will last for years with 24x7 use. WD Red drives are designed for use in NAS devices and have nice long trouble free lives. At least, the 8TB models do. Have not tried any of the newer 14TB models. 

 

Note the difference between WD Red and WD Red Pro. Same goes for Seagate IronWolf and IronWolf Pro.

 

I use only the Red Pro and IronWolf Pro which have two years longer warranty.

[Miska]

6 hours ago, mevdinc said:

Can anyone give me info on the possible customs duty on the Terminator DAC shipped to an address in the UK or EU?
Many thanks.

 

Likely only notable one is your local VAT. Customs for electronics is usually only few percents and thus not so much compared to VAT.

 

[Miska]

On 6/16/2019 at 6:08 PM, StreamFidelity said:

6bit DSD hardware decoding with 32 steps FIR filters

 

Does DSD not always run with 1bit?

 

R-2R decoding is exclusively for PCM? With which hardware is DSD decoded then?

 

That sounds like my DSC1 DAC design. For SDM (including DSD) DACs, number of bits is sort of irrelevant. For that reason usually number of levels is used instead.

 

Yes, R2R is for PCM and separate 32-element (33-level) converter is for DSD.

 

[Miska]

14 hours ago, StreamFidelity said:

Loud crackling when played for the first time.
Soft crackling on a new stream like from a turntable.

 

Does anyone know that? For a few weeks now I've been going directly fom Terminator to my T + A M10 power amplifiers via xlr. The sound is great and better than over preamp. The volume is controlled via HQPlayer. It is played via USB from the audio PC.

 

I've only heard the crackle since bypassing the preamp. They were probably always there, but dejected by preamp.

 

Small DC offset in the DAC output, possibly output combined with mute relay? May be handled differently by a preamp DC block/servo than by a power amp, and at least attenuated by the volume control.

 

This is not at all unusual behavior for a DC-connected DAC output stage. Other alternative is a series capacitor in DAC output that is floating until actually gets connected due to DAC unmuting the output. So it's DC-block behavior only begins to settle after first unmute. Time constant can be like 1 Hz or 0.1 Hz, so it takes some time to settle.

 

[Miska]

And good devices convert single ended analog output from differential balanced signal with a proper conversion.

 

While simple external adapters pick just one half of the differential signal and leaves the other one unused. Unfortunately some devices also internally create single ended output from the balanced this way (cheaper).

 

[Miska]

9 hours ago, asdf1000 said:

Hi @Miska

 

Can you circle ⭕️ where the DSD is converted to analogue?

 

I imagine most of this board (resistors) is devoted to converting PCM to analogue but confused where DSD action is

 

 

Could be the array in the middle, and PCM on the sides.

 

But also with suitable board and circuit design you could also reconfigure R2R ladder into AFIR converter. It needs different connection arrangement but same parts could be used.